Harmful meme detection on Harm-C (test)
87AccuracyEXPLAINHM
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| EXPLAINHM2024.01 | 87 | 86.41 | |
| Pro-Cap2024.01 | 85.01 | 83.17 | |
| MaskPrompt2024.01 | 84.47 | 81.51 | |
| MOMENTA2024.01 | 83.82 | 82.8 | |
| VisualBERT COCOPre-training dataset=COCO2024.01 | 81.36 | 80.13 | |
| VILBERT CCPre-training dataset=CC2024.01 | 78.7 | 78.09 | |
| MMBT2024.01 | 73.48 | 67.12 | |
| Late Fusion2024.01 | 73.24 | 70.25 | |
| Text BERT2024.01 | 70.17 | 66.25 | |
| Image-Region2024.01 | 68.74 | 62.97 |